.p2p top-level domain

[bascule]

So, in the wake of the US government seizing the domain names of several p2p file sharing sites, people behind some of the seized domains including Torrent-Finder are proposing a new .p2p top-level domain which would be powered by BitTorrent itself. Some are further suggesting that the .p2p domain registry itself be distributed.

As a general P2P enthusiast, I have to say: this is dumb. I'm sure most of you can see why immediately, but what's being violated here is Zooko's Triangle, a set of rules for identity systems, specifically: decentralized, secure, human-meaningful... pick two.

There exists no secure way to provide a decentralized registry of human meaningful names. Someone much be in charge of the registry, and as soon as that happens there's someone worldwide legal systems can target. Otherwise, you create a decentralized system which is subject to forgery and other types of attacks.

While I think the idea of a centralized, non-ICANN controlled .p2p domain is cool, it only works until the people behind it are arrested, drug into court, and jailed, much like what just happened to the founders of The Pirate Bay.

[kcdclan]

Re: .p2p top-level domain

I do not see how you can tag a search engine responsible.
They are like the friend who knows a drug dealer... there not really at fault... they just point you in that direction.
Kind of like saying lets arrest anyone who talks about it..
I mean really... seems really stupid to me..

[bascule]

Re: .p2p top-level domain

I'm sure everyone here has seen discussion around SOPA. SOPA would give the US government widespread power to block certain domains.

This makes the concept of a fully decentralized domain system pretty interesting. This is a very difficult problem to solve, and that problem has a name, Zooko's triangle:

How do you build a system which is secure, decentralized, and human-meaningful?

We need secure human-meaningful domain names, but the present system, while "secure" (for lack of widespread DNSSEC adoption) is completely centralized. This lets governments terminate domains at will.

A particularly interesting solution to this problem is Namecoin. Namecoin uses the same proof-of-work system from BitCoin to provide distributed DNS. One can purchase domain names in exchange for solving proof-of-work problems, and they're entered into a BitCoin-like block chain.

The basic idea behind block chains is that peers in the system try to solve a complex proof-of-work problem which can only be solved in a brute force computationally complex
manner. Once a particular node happens on a solution, it can send it to the network, which will gossip that solution to other nodes. Every node accepts the longest copy of the block chain, so the system naturally favors cooperation.
They're using the domain name .bit, and DNS servers are available which will resolve domains using both the normal DNS system and .bit domains, or you can run your own.

[The Dark Tangent]

Re: .p2p top-level domain

I met the author of namecoin in Tokyo, he is really passionate about this issue.

Does anyone track all the various proposals and projects?

There is one or two groups wanting to replace the root with alternate roots, a couple p2p DNS replacements, as well as overlay ideas like using .onion for everything.

[bascule]

Re: .p2p top-level domain

Does anyone track all the various proposals and projects?

Here's a blog post covering some of the projects in this space:

http://changaco.net/blog/DNS_problems_and_alternatives/

One I hadn't heard about before is CoDoNS, which seeks to work as a sort of P2P "safety net" for the DNS system, whether names can't be resolved because they've been forcibly removed from the registry, or even if someone were to pull off a successful DDoS against the existing DNS system. CoDoNS looks particularly interesting because it's fully integrated with the existing DNS system and DNSSEC:

http://www.cs.cornell.edu/people/egs/beehive/codons.php